Purpose. Our main goal is to quantify the impact of a low-fat high-fiber (LFHF) diet and/or a soy supplement on hormonal patterns associated with breast cancer (BC) risk. A secondary purpose is to assess the convenience and precision of using daily saliva samples to measure ovarian steroids throughout the menstrual cycle. Background/Significance. The hypothesis that excess cumulative exposure to estradiol (E2) and progesterone (PG) in premenopausal women increases the risk of developing BC offers a unifying explanation for diverse epidemiologic and laboratory findings. A second hypothesis, relating BC to overproduction of 16alpha vs. 2-hydroxylated E2 metabolites, is supported by a growing body of evidence. The LFHF diet has been linked with reduced E2 levels and decreased 16alpha-hydroxylation in premenopausal women, but the evidence is sparse and inconsistent. A large randomized trial, featuring a minimal intervention comparison group, longer follow-up, and more reliable measures of hormone levels in cycling women can contribute valuable evidence on dietary control of BC risk through hormonal mechanisms. Soy foods containing high levels of phytoestrogens might also have a favorable effect on hormone profiles. Phytoestrogen intake has been linked to an increase in hepatic production of SHBG and a resultant decrease in free E2. We propose a randomized placebo-controlled trial of soy supplement to evaluate its effects both alone and in combination with the LFHF diet. Salivary measurements of E2 and PG offer the possibility of far richer and more sensitive datasets for studying the impact of dietary factors on time-varying levels of bioavailable hormone, but have never been evaluated in an intervention setting. Methods. 200 women (non-obese, regular periods, aged 30-45) will follow either usual or LFHF diet and will provide hormonal and other data at baseline (B), 4 and 12 mos. At 12 mos, women in both groups will be re- randomized to either soy supplement or placebo and a final follow-up will occur at 16 mos. (Details of the diet intervention in the accompanying IRPG proposal.) In addition to keeping menstrual diaries, 140 women will provide at B, 4, 12 and 16 mos.: i) one midluteal blood sample for E2, PG and SHBG levels, and ii) midluteal urine for l6alpha and 2-OHE1 assays. A Frequent Sample (FS) subgroup of 60 women will provide the above plus, iii) blood samples on day 7 and every 3rd day from day 14 to next menses onset (6-7 samples/cycle), and iv) daily saliva samples for a full cycle at B, 4, 12 and l6mos. The main analysis will compare mean changes in serum E2, PG and SHBG and urine l6alpha:2OHE1 ratios across diet groups. In the FS subgroup, we will use salivary PG to select the best-timed bloods for midluteal analysis and compare cumulative salivary E2 and PG across diet groups. Blood, urine, breast fluid and mammograms (in women over age 40) for digitized parenchymal density readings will be banked for separately funded analyses.